Flow Matching-Based PET Image Reconstruction

Fumio Hashimoto; Ziqian Huang; Tatsuya Yokota; Kuang Gong

eess.IVAug 20, 2026

Flow Matching-Based PET Image Reconstruction

Authors: Fumio Hashimoto, Ziqian Huang, Tatsuya Yokota, Kuang Gong

Organizations: J. Crayton Pruitt Family Department of Biomedical Engineering, University of Florida, FL, USA · Department of Computer Science, Nagoya Institute of Technology, Aichi, Japan · RIKEN Center for Advanced Intelligence Project, Tokyo, Japan

Abstract

Generative models have shown strong potential for positron emission tomography (PET) image reconstruction. Although diffusion model-based reconstruction methods have demonstrated promising performance, they often require many reverse sampling steps with data-consistency updates incorporated into the sampling process. Flow matching offers an attractive alternative because it can directly estimate clean images from intermediate states, allowing data-consistency refinement to be separated from flow propagation. In this work, we proposed flow matching-based PET image reconstruction methods. We first established PET-FlowDPS by incorporating Poisson likelihood guidance with an expectation-maximization (EM)-based preconditioner into the FlowDPS framework. We then proposed a model-based PET reconstruction method that used a pretrained flow matching model as a prior, in which the flow-based prior, PET data refinement, and stochastic propagation were interpreted within an approximate Bayesian framework. Experimental results using [ $^{\text{18}}\text{F}$ ]FDG brain PET datasets showed that the proposed method achieved better bias-variance trade-offs across different dose levels compared with other reference methods. These results demonstrated the potential of flow matching as a generative prior for quantitative PET image reconstruction.

Figures & tables

Fig. 1: Overview of (a) PET-FlowDPS and (b) the proposed model-based PET reconstruction with a flow-based prior. In step (1), both methods estimate a destination image using the pretrained flow model. In step (2), (a) PET-FlowDPS applies a data-consistency update using the gradient of the log-likelihood, and (b) the proposed method performs a MAP reconstruction that balances data fidelity and the conditional flow-based prior. In step (3), the refined destination estimate is propagated to the next flow time point using the respective stochastic propagation rules. The purple contours in step (2) represent the PET log-likelihood, and the blue contours in (b-2) represent the conditional flow-based prior approximated by a Gaussian distribution.

Fig. 2: Impact of the number of sampling steps

K

on PSNR (top) and on the putamen uptake–white matter CoV tradeoff curves (bottom). Filled markers correspond to the images as shown in Fig. 4 .

Fig. 3: Impact of the number of inner EM iterations

N

on the putamen uptake–white matter CoV trade-off curves across the 10 subjects at the 10% count level. Markers were plotted for

N=1,2,\ldots,14

. Filled markers indicate

N=10

, which was used for the representative images shown in Fig. 4 .

Fig. 4: Reconstruction results of three orthogonal slices of the clinical [ ${}^{\text{18}}\text{F}$ ]FDG data at 10% dose.

Fig. 5: Mean putamen uptake–white matter CoV trade-off curves across the 10 subjects at the 10% count level. Markers were plotted at 10-iteration intervals from 10 to 100 for MLEM, at 20-iteration intervals from 20 to 200 for MAPEM (

\beta_{\mathrm{MAPEM}}=30

\gamma=2

), at

\lambda_{\text{DDS}}

of 0, 10, 100, 500, and 1000 for DDS, at

\lambda

of 0.8, 0.9, 1.0, 1.2, and 1.4 for PET-FlowDPS, and at

\beta

of 0.05, 0.06, 0.08, 0.1, and 0.25 for Proposed and Proposed(5). Filled markers correspond to the representative images shown in Fig. 4 .

Fig. 6: Voxel-wise mean, bias, and standard-deviation maps calculated from 10 independent 10% low dose realizations for the different reconstruction methods.

Fig. 7: Reconstruction results of three orthogonal slices of the clinical [ ${}^{\text{18}}\text{F}$ ]FDG data at 5% dose.

Fig. 8: Mean putamen uptake–white matter CoV trade-off curves across the 10 subjects at the 5% count level. Markers were plotted at 10-iteration intervals from 10 to 100 for MLEM, at 10-iteration intervals from 10 to 200 for MAPEM, at

\lambda_{\text{DDS}}

of 0, 10, 100, 500, and 1000 for DDS, at

\lambda

of 0.8, 0.9, 1.0, 1.2, 1.4, 1.6, 1.8, 2.0, 2.2, and 2.5 for PET-FlowDPS, and at

\beta

of 0.005, 0.01, 0.02, 0.03, 0.04, 0.05, 0.06, 0.08, 0.1, and 0.25 for Proposed and Proposed(5). Filled markers correspond to the images as shown in Fig. 7 .

Explore similar work

Physics-Guided Flow Matching for CT Image Reconstruction
Aug 28, 2026Davide EvangelistaImage Reconstruction Computed Tomography
Department of Computer Science and Engineering University of Bologna Bologna, 40126, Italy
FlowPET: Physics-Informed Symplectic Flow Matching for Low-Count PET Reconstruction
Jul 13, 2026Zheng Zhang, Hao Tang, Yingying Hu +2Flow Matching Image Reconstruction
Centre for Smart Health, The Hong Kong Polytechnic University, Hong Kong, China · Department of Nuclear Medicine, Sun Yat-sen University Cancer Center, Guangzhou, China · Research Center for Medical AI, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China
Efficient Flow Matching for Sparse-View CT Reconstruction
Feb 27, 2026Jiayang Shi, Lincen Yang, Zhong Li +3Flow Matching Sparse-View CT Reconstruction
Centrum Wiskunde en Informatica · LIACS, Leiden University · Great Bay University +1

Aug 28, 2026J/K move · Enter open · S save

Physics-Guided Flow Matching for CT Image Reconstruction

Davide Evangelista

Department of Computer Science and Engineering University of Bologna Bologna, 40126, Italy

Image Reconstruction Computed Tomography Flow Matching Inverse Problems Diffusion-Based Inverse Problems

Open paper AskPDF arXiv

Deep generative models have recently emerged as powerful priors for solving ill-posed inverse problems in CT, with diffusion-based approaches achieving state-of-the-art reconstruction performance. However, diffusion models typically rely on stochastic sampling procedures, long inference trajectories, and carefully tuned noise schedules, which can limit computational efficiency and numerical stability, especially at high spatial resolutions. In this work, we investigate Flow Matching as an alternative generative prior for CT reconstruction. We train a high-resolution Rectified Flow Matching model on 256x256 chest images from the Mayo Clinic Low-Dose CT dataset. To mitigate overfitting and limited anatomical variability, we employ a two-stage training strategy consisting of an initial phase with strong, anatomically informed data augmentation, followed by a fine-tuning phase with reduced or no augmentation to refine structural fidelity. The resulting model is capable of generating high-quality and anatomically coherent CT-like images, serving as a strong learned prior. We then evaluate multiple reconstruction methods specifically designed for Flow Matching models, including Plug-and-Play Flow, FlowDPS, Flower, and Flow-Priors (ICTM), and compare them against state-of-the-art diffusion-based reconstruction algorithms such as DDRM, DPS, and DiffPIR. Experimental results across several CT inverse problem settings show that Flow Matching-based approaches consistently outperform diffusion-based methods in terms of PSNR, SSIM, and perceptual quality, while requiring fewer sampling steps. Finally, we publicly release the trained Flow Matching model and accompanying code to facilitate reproducibility and future research. Overall, this work demonstrates that Flow Matching provides a stable, efficient, and effective alternative to diffusion models for high-resolution CT image reconstruction.

Abstract

Explore similar work

Physics-Guided Flow Matching for CT Image Reconstruction

FlowPET: Physics-Informed Symplectic Flow Matching for Low-Count PET Reconstruction

Efficient Flow Matching for Sparse-View CT Reconstruction